diff --git a/src/ocrmypdf/hocr.py b/src/ocrmypdf/hocr.py
new file mode 100644
index 00000000..ff177a65
--- /dev/null
+++ b/src/ocrmypdf/hocr.py
@@ -0,0 +1,266 @@
+from __future__ import annotations
+
+import importlib
+import re
+from io import BufferedReader, BufferedWriter, BytesIO
+from pathlib import Path
+
+import pikepdf
+from bs4 import BeautifulSoup
+from pikepdf import (
+    ContentStreamInstruction,
+    Dictionary,
+    Name,
+    Operator,
+    Pdf,
+    unparse_content_stream,
+)
+
+GLYPHLESS_FONT = importlib.resources.read_binary("ocrmypdf", "pdf.ttf")
+CHAR_ASPECT = 2
+
+
+def parse_bbox(title):
+    # Match for bbox pattern
+    bbox_pattern = re.compile(r'bbox (\d+) (\d+) (\d+) (\d+)')
+    match = bbox_pattern.search(title)
+    if match:
+        return tuple(map(int, match.groups()))
+    else:
+        return None
+
+
+def register_glyphlessfont(pdf: Pdf):
+    """Register the glyphless font.
+
+    Create several data structures in the Pdf to describe the font. While it create
+    the data, a reference should be set in at least one page's /Resources dictionary
+    to retain the font in the output PDF and ensure it is usable on that page.
+    """
+    PLACEHOLDER = Name.Placeholder
+
+    basefont = pdf.make_indirect(
+        Dictionary(
+            BaseFont=Name.GlyphLessFont,
+            DescendantFonts=[PLACEHOLDER],
+            Encoding=Name("/Identity-H"),
+            Subtype=Name.Type0,
+            ToUnicode=PLACEHOLDER,
+            Type=Name.Font,
+        )
+    )
+    cid_font_type2 = pdf.make_indirect(
+        Dictionary(
+            BaseFont=Name.GlyphLessFont,
+            CIDToGIDMap=PLACEHOLDER,
+            CIDSystemInfo=Dictionary(
+                Ordering="Identity",
+                Registry="Adobe",
+                Supplement=0,
+            ),
+            FontDescriptor=PLACEHOLDER,
+            Subtype=Name.CIDFontType2,
+            Type=Name.Font,
+            DW=1000 // CHAR_ASPECT,
+        )
+    )
+    basefont.DescendantFonts = [cid_font_type2]
+    cid_font_type2.CIDToGIDMap = pdf.make_stream(b"\x00\x01" * 65536)
+    basefont.ToUnicode = pdf.make_stream(
+        b"/CIDInit /ProcSet findresource begin\n"
+        b"12 dict begin\n"
+        b"begincmap\n"
+        b"/CIDSystemInfo\n"
+        b"<<\n"
+        b"  /Registry (Adobe)\n"
+        b"  /Ordering (UCS)\n"
+        b"  /Supplement 0\n"
+        b">> def\n"
+        b"/CMapName /Adobe-Identify-UCS def\n"
+        b"/CMapType 2 def\n"
+        b"1 begincodespacerange\n"
+        b"<0000> <FFFF>\n"
+        b"endcodespacerange\n"
+        b"1 beginbfrange\n"
+        b"<0000> <FFFF> <0000>\n"
+        b"endbfrange\n"
+        b"endcmap\n"
+        b"CMapName currentdict /CMap defineresource pop\n"
+        b"end\n"
+        b"end\n"
+    )
+    font_descriptor = pdf.make_indirect(
+        Dictionary(
+            Ascent=1000,
+            CapHeight=1000,
+            Descent=-1,
+            Flags=5,  # Fixed pitch and symbolic
+            FontBBox=[0, 0, 1000 // CHAR_ASPECT, 1000],
+            FontFile2=PLACEHOLDER,
+            FontName=Name.GlyphLessFont,
+            ItalicAngle=0,
+            StemV=80,
+            Type=Name.FontDescriptor,
+        )
+    )
+    font_descriptor.FontFile2 = pdf.make_stream(GLYPHLESS_FONT)
+    cid_font_type2.FontDescriptor = font_descriptor
+    return basefont
+
+
+class ContentStreamSequence:
+    def __init__(self, instructions=None):
+        self._instructions: list[ContentStreamInstruction] = instructions or []
+
+    def push(self):
+        """Save the graphics state."""
+        inst = [ContentStreamInstruction([], Operator("q"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def pop(self):
+        """Restore the graphics state."""
+        inst = [ContentStreamInstruction([], Operator("Q"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def cm(self, a: float, b: float, c: float, d: float, e: float, f: float):
+        """Concatenate matrix."""
+        inst = [ContentStreamInstruction([a, b, c, d, e, f], Operator("cm"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def begin_text(self):
+        """Begin text object."""
+        inst = [ContentStreamInstruction([], Operator("BT"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def end_text(self):
+        """End text object."""
+        inst = [ContentStreamInstruction([], Operator("ET"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def begin_marked_content(self, mctype: Name, mcid: int):
+        """Begin marked content sequence."""
+        inst = [
+            ContentStreamInstruction([mctype, Dictionary(MCID=mcid)], Operator("BDC"))
+        ]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def end_marked_content(self):
+        """End marked content sequence."""
+        inst = [ContentStreamInstruction([], Operator("EMC"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def set_text_font(self, font: Name, size: int):
+        """Set text font and size."""
+        inst = [ContentStreamInstruction([font, size], Operator("Tf"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def set_text_matrix(
+        self, a: float, b: float, c: float, d: float, e: float, f: float
+    ):
+        """Set text matrix."""
+        inst = [ContentStreamInstruction([a, b, c, d, e, f], Operator("Tm"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def set_text_rendering(self, mode: int):
+        """Set text rendering mode."""
+        inst = [ContentStreamInstruction([mode], Operator("Tr"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def set_text_horizontal_scaling(self, scale: float):
+        """Set text horizontal scaling."""
+        inst = [ContentStreamInstruction([scale], Operator("Tz"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def show_text(self, text: str):
+        """Show text."""
+        inst = [ContentStreamInstruction([[text.encode("utf-16be")]], Operator("TJ"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def stroke_and_close(self):
+        """Stroke and close path."""
+        inst = [ContentStreamInstruction([], Operator("s"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def append_rectangle(self, x: float, y: float, w: float, h: float):
+        """Append rectangle to path."""
+        inst = [ContentStreamInstruction([x, y, w, h], Operator("re"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+    def set_stroke_color(self, r: float, g: float, b: float):
+        """Set RGB stroke color."""
+        inst = [ContentStreamInstruction([r, g, b], Operator("RG"))]
+        return ContentStreamSequence(self._instructions + inst)
+
+
+class ContentStreamBuilder:
+    def __init__(self):
+        self._instructions = []
+
+    def build(self):
+        return self._instructions
+
+    def add(self, other: ContentStreamSequence):
+        self._instructions.extend(other._instructions)
+
+
+def hocr_to_pdf(hocr_stream: BufferedReader, output_stream: BufferedWriter):
+    # Parse the hOCR data
+    soup = BeautifulSoup(hocr_stream, 'lxml')
+
+    # Find the page size from the hOCR input
+    ocr_page = soup.find('div', class_='ocr_page')
+    if not ocr_page or 'title' not in ocr_page.attrs:
+        raise ValueError("hOCR input does not contain page information")
+    page_bbox = parse_bbox(ocr_page['title'])
+    if page_bbox is None:
+        raise ValueError("Could not parse page bounding box from hOCR input")
+    _, _, page_width, page_height = page_bbox
+
+    # Create a new PDF with pikepdf
+    pdf = pikepdf.new()
+    page = pdf.add_blank_page(page_size=(page_width, page_height))
+
+    font_name = Name("/f-0-0")
+
+    page.Resources = Dictionary(
+        Font=Dictionary({font_name: register_glyphlessfont(pdf)})
+    )
+
+    cs = ContentStreamBuilder()
+    cs.add(ContentStreamSequence().push().cm(1, 0, 0, -1, 0, page_height))
+
+    # Add content using these fonts
+    for span in soup.find_all('span', class_='ocrx_word'):
+        if 'title' not in span.attrs:
+            continue
+        word_bbox = parse_bbox(span['title'])
+        if not word_bbox:
+            continue
+        x0, y0, x1, y1 = word_bbox
+        text = span.get_text() if span.get_text() else ''
+
+        cos_a, sin_a = 1, 0
+        font_size = y1 - y0
+        space_width = 0
+        box_width = x1 - x0 + space_width
+
+        h_stretch = 100.0 * box_width / len(text) / font_size * CHAR_ASPECT
+
+        cs.add(
+            ContentStreamSequence()
+            .begin_text()
+            .set_text_rendering(3)
+            .set_text_matrix(cos_a, -sin_a, sin_a, cos_a, x0, y1)
+            .set_text_font(font_name, font_size)
+            .set_text_horizontal_scaling(h_stretch)
+            .show_text(text)
+            .end_text()
+        )
+
+    cs.add(ContentStreamSequence().pop())
+
+    page.Contents = pdf.make_stream(unparse_content_stream(cs.build()))
+
+    # Save the PDF to a file or return as a byte string
+    pdf.save(output_stream)
+    pdf.close()
diff --git a/src/ocrmypdf/hocrtransform/__init__.py b/src/ocrmypdf/hocrtransform/__init__.py
index 711a471c..5f30c9d9 100755
--- a/src/ocrmypdf/hocrtransform/__init__.py
+++ b/src/ocrmypdf/hocrtransform/__init__.py
@@ -11,21 +11,19 @@ from __future__ import annotations
 import argparse
 import os
 import re
-import warnings
 from math import atan, cos, sin
 from pathlib import Path
 from typing import Any, NamedTuple
 from xml.etree import ElementTree
 
-with warnings.catch_warnings():
-    # reportlab uses deprecated load_module
-    # shim can be removed when we require reportlab >= 3.7
-    warnings.filterwarnings(
-        'ignore', category=DeprecationWarning, message=r".*load_module.*"
-    )
-    from reportlab.lib.colors import black, cyan, magenta, red
-    from reportlab.lib.units import inch
-    from reportlab.pdfgen.canvas import Canvas
+from ocrmypdf.hocrtransform.backends.reportlab import (
+    Canvas,
+    black,
+    cyan,
+    inch,
+    magenta,
+    red,
+)
 
 # According to Wikipedia these languages are supported in the ISO-8859-1 character
 # set, meaning reportlab can generate them and they are compatible with hocr,
diff --git a/src/ocrmypdf/hocrtransform/backends/__init__.py b/src/ocrmypdf/hocrtransform/backends/__init__.py
new file mode 100644
index 00000000..e69de29b
diff --git a/src/ocrmypdf/hocrtransform/backends/reportlab.py b/src/ocrmypdf/hocrtransform/backends/reportlab.py
new file mode 100644
index 00000000..e9b1fc2d
--- /dev/null
+++ b/src/ocrmypdf/hocrtransform/backends/reportlab.py
@@ -0,0 +1,13 @@
+from __future__ import annotations
+
+import warnings
+
+with warnings.catch_warnings():
+    # reportlab uses deprecated load_module
+    # shim can be removed when we require reportlab >= 3.7
+    warnings.filterwarnings(
+        'ignore', category=DeprecationWarning, message=r".*load_module.*"
+    )
+    from reportlab.lib.colors import black, cyan, magenta, red
+    from reportlab.lib.units import inch
+    from reportlab.pdfgen.canvas import Canvas